app.py

import gradio as gr
from transformers import pipeline

# Load small + free LLM
generator = pipeline("text2text-generation", model="google/flan-t5-small")

# --- Agent 1: Summarizer + Knowledge Graph Builder ---
def doc_agent(user_text):
    # Step 1: Summarize
    summary_prompt = f"Summarize this in 3 lines: {user_text}"
    summary = generator(summary_prompt, max_length=80, do_sample=False)[0]['generated_text']

    # Step 2: Build a "mini knowledge graph" (keywords as nodes)
    keyword_prompt = f"Extract 5 important keywords from this text: {user_text}"
    keywords = generator(keyword_prompt, max_length=40, do_sample=False)[0]['generated_text']
    graph_nodes = [kw.strip() for kw in keywords.split(",") if kw.strip()]
    graph_repr = " → ".join(graph_nodes) if graph_nodes else "No graph generated."

    return f"📄 Summary:\n{summary}\n\n🕸️ Knowledge Graph:\n{graph_repr}"

# --- Agent 2: Career Skill Recommender ---
def career_agent(user_goal):
    # Step 1: Analyze career intent
    analysis_prompt = f"Identify skill gap for this career goal: {user_goal}"
    analysis = generator(analysis_prompt, max_length=50, do_sample=False)[0]['generated_text']

    # Step 2: Recommend roadmap
    roadmap_prompt = f"Suggest a 3-step learning roadmap for: {user_goal}"
    roadmap = generator(roadmap_prompt, max_length=80, do_sample=False)[0]['generated_text']

    return f"🔍 Gap Analysis:\n{analysis}\n\n🛠️ Skill Roadmap:\n{roadmap}"

# --- Combined Agent Controller ---
def agentic_ai(user_input, mode):
    if mode == "Document Insight":
        return doc_agent(user_input)
    elif mode == "Career Roadmap":
        return career_agent(user_input)
    else:
        return "⚠️ Please choose a valid mode."

# --- Demo UI ---
demo = gr.Interface(
    fn=agentic_ai,
    inputs=[
        gr.Textbox(lines=4, placeholder="Enter text or career goal..."),
        gr.Radio(["Document Insight", "Career Roadmap"], label="Choose Mode")
    ],
    outputs="text",
    title="🌐 Mini Agentic AI MVP",
    description="""
    This smallest MVP demonstrates:
    - 📄 Document Summarization
    - 🕸️ Knowledge Graph (mini keyword graph)
    - 🧑‍💻 Career Skill Gap Analysis
    - 🛠️ Personalized 3-step Roadmap

    Built with free Hugging Face + Gradio. Optimized for AI Research use cases.
    """
)

if __name__ == "__main__":
    demo.launch()